Australians are sitting on one of the largest pools of retirement savings in the world. As of 30 June 2025, Australians held AU$4.33 trillion in superannuation assets, making Australia the fourth-largest holder of pension fund assets globally. Analysis projects that Australian pension assets will surpass those of the United Kingdom in 2030 and those of Canada in 2031, rising to second in the world. Against that backdrop, the question of how retirees convert those savings into sustainable income has never been more important. 

Yet the evidence suggests that most Australians are not doing this well. The Association of Superannuation Funds of Australia estimates that couples need to save AU$690,000 and singles AU$595,000 by age 67 for a comfortable retirement – yet the average Australian retires with significantly less, and even those who do reach that threshold routinely spend and draw down less than they should.

The retirement income problem is not just an Australian one. Researchers around the world have struggled for decades to explain three persistent and related puzzles of retirement behaviour. Retirees consistently buy fewer retirement income products (such as life annuities, known as longevity insurance) than economic models predict they should. They also spend noticeably less of their savings than their resources allow. And they draw down their accumulated wealth far more slowly than expected. These patterns hold across multiple countries and income levels, and they have defied easy explanation. 

Learn more: One size doesn't fit all: Is the super system failing vulnerable members?

In Australia, converting a lump sum into a reliable income stream remains a significant concern for retirees, with 65% of Australians expressing concern about not knowing how much they need to save for retirement to meet their income needs – a striking figure in a country with a compulsory superannuation system that has been operating for more than 30 years.

Now, new research from an international team of mathematicians and actuaries offers an explanation for all three puzzles at once. The answer, it turns out, may lie not in irrational behaviour or poor financial literacy, but in something far more mundane: the fixed fee that retirees pay every time they buy additional annuity income. That single, overlooked cost (the flat upfront charge triggered each time a retiree approaches an insurer) turns out to reshape retirement decision-making in ways that are rational, predictable, and ultimately damaging to retirement outcomes. With an ageing population, understanding how it works and what to do about it is more important than ever.

“The research was motivated by the disconnection between theory and practice: many retirement models assume people can buy annuities smoothly, whenever they want, and without extra costs,” said Yang Shen, an Associate Professor in the School of Risk and Actuarial Studies at UNSW Business School and an Associate Investigator at the UNSW Centre for Population Ageing Research (CEPAR).

“But real life is not that simple. Annuity purchases usually happen in steps, and each step may involve fees, paperwork, advice costs or other hurdles. The research explored the impact of fixed annuity costs on retirement decisions and outcomes.”

The true cost of buying an annuity

A life annuity is a financial product that converts a lump sum of savings into a regular income stream for the rest of a person's life (essentially, a private pension). Economic theory has long suggested that rational retirees should convert most of their wealth into annuity income, because doing so removes the risk of outliving their savings. Yet in practice, voluntary annuity take-up rates remain persistently low. Researchers call this the "annuity puzzle". It sits alongside two other well-documented mysteries of retirement behaviour: the 'retirement-consumption puzzle', in which retirees spend noticeably less than their resources would allow; and the 'retirement-savings puzzle', in which retirees draw down their accumulated wealth far more slowly than expected.

Most people think of annuity costs in terms of the purchase price itself: the lump sum paid to secure a regular income. But every time a retiree approaches an insurer to buy additional annuity income, they typically face a fixed administrative fee. Unlike proportional costs, which rise and fall with the size of the transaction, a fixed fee applies regardless of the amount of annuity income purchased. This means that buying annuities in small amounts or frequently is economically punishing, as the fee eats into the value of each small purchase.

Subscribe to BusinessThink for the latest research, analysis and insights from UNSW Business School

This issue sits at the centre of a new study published in Insurance: Mathematics and Economics, one of the field's leading actuarial journals. The paper, Optimal annuitisation and asset allocation with fixed transaction costs, was coauthored by Professor Junyi Guo at Nankai University, Associate Professor Xiaoqing Liang at the Hebei University of Technology, UNSW Business School’s A/Prof. Yang Shen and Professor Jie Xiong at the Southern University of Science and Technology.

Using mathematical techniques drawn from stochastic control theory and duality methods (tools designed to find the best possible decision-making strategy over time, given uncertainty) the researchers modelled the optimal behaviour of a retiree who can invest in both a risky asset (such as shares) and a risk-free asset (such as government bonds), and who can also purchase life annuities, subject to a fixed transaction fee each time.

The researchers found that fixed costs make retirees less likely to buy annuities in small amounts. Instead, A/Prof. Shen suggested it is better to wait until there is enough wealth to absorb the cost, then make a larger purchase. “In simple terms, the fixed cost changes annuity buying from a gradual process into a ’wait, then act’ decision,” he said.

“What is most interesting is that the impact goes beyond annuity purchases. The cost also affects how much retirees spend, how much they keep in savings, and how quickly they draw down their wealth. This means that behaviour which may look overly cautious at first does make sense once these real-world costs are taken into account.”

A wealth threshold that changes everything

The central finding of the research is that the optimal strategy for managing annuity purchases follows a "two-barrier" rule, as mathematicians call it. In plain English, a retiree should do nothing about buying more annuity income until their wealth rises to a certain upper threshold. Once wealth crosses that threshold, they should purchase enough annuity income to bring their wealth back down to a lower threshold. Then the process repeats.

Without any transaction costs, this strategy collapses to a single threshold, and retirees would buy annuities more frequently and in smaller amounts. But once a fixed fee is introduced, the upper threshold rises substantially while the lower threshold falls, widening the gap between the two. In numerical examples run by the researchers, calibrated to parameters reflecting a typical 65-year-old male retiree in the United States, the upper wealth barrier in the transaction-cost model was approximately 225.94 units, compared to a single barrier of 51.26 units in the no-transaction-cost scenario. The lower barrier in the transaction-cost model sat at 26.23 units. These numbers illustrate just how dramatically even a modest fixed cost reshapes optimal annuitisation decisions.

Learn more: Income is key in scramble to fix retirement phase of super

The researchers also found that when transaction costs increase, retirees should purchase larger amounts of annuity income each time they do buy, rather than making small, frequent purchases. Conversely, as transaction costs fall toward zero, the two barriers converge to the single barrier that characterises the frictionless ideal.

How fees reshape spending and savings behaviour

The implications of this two-barrier dynamic extend well beyond annuity purchasing decisions. Through simulations involving 10,000 modelled retirement trajectories, the researchers demonstrated that fixed transaction costs produce a cascade of downstream effects on how retirees invest, spend, and save.

Under the transaction-cost model, retirees tend to hold onto more wealth for longer because they are waiting to accumulate enough to make a meaningful annuity purchase worthwhile. Their wealth initially rises after retirement, then declines, a pattern that contrasts with the steady drawdown seen in the no-transaction-cost scenario. The paper found that transaction costs not only delayed retirees' decisions to annuitise additional wealth but also led to underspending and a slower drawdown rate in the decumulation phase.

The research also found that transaction costs produced a distinctive pattern in consumption: retirees spend less than they otherwise could, particularly in the early years of retirement. Their consumption dips, recovers, and stabilises at a higher level later in retirement as annuity income accumulates. This pattern is consistent with the retirement-consumption puzzle, where retirees persistently spend below their means in ways that classical economic models do not predict.

What this means for retirees and financial advisers

The research carries tangible implications for individuals, financial advisers, and policymakers. For practitioners advising retirees on decumulation strategies (how to draw down accumulated wealth in retirement), the findings suggest that fixed transaction costs are not a minor administrative inconvenience. They are a structurally significant friction that distorts optimal decisions in ways that compound over time.

The research indicates that the point at which a retiree should purchase annuity income shifts significantly depending on annuity prices, the retiree's age and health outlook, market volatility, and individual risk aversion. Retirees with higher risk aversion, those more uncomfortable with financial uncertainty, face lower optimal purchase thresholds, meaning they should move toward annuitisation earlier and more aggressively. Conversely, as retirees age and the subjective probability of death rises, the wealth thresholds at which annuitisation makes sense also rise, suggesting that older retirees may rationally delay further annuity purchases.

Learn more: The new challenges reshaping Australia’s retirement system

For financial advisers and product designers, the findings point to the value of reducing or restructuring fixed transaction fees. Even small reductions in per-transaction costs produce meaningful shifts in the optimal purchase threshold, encouraging earlier and more frequent annuitisation. Policy makers seeking to improve retirement outcomes might also consider whether regulatory or market-design reforms could reduce the friction costs that currently suppress voluntary annuity take-up.

A/Prof. Shen explained that the main takeaway from the research is that small frictions can make a big difference in retirement. “A fixed cost may seem like a minor product feature, but it can change when retirees buy annuities, how much they buy, and how confident they feel about spending their savings. A joint effort of advisers, providers and policymakers is needed to make annuity decisions more flexible, affordable and better timed, so that retirees can turn their savings into a steady income more effectively,” he concluded.